An siRNA capable of suppressing the expression of a specific gene is widely used as a research tool. An siRNA is also receiving attention for applications to therapeutic agents for a wide variety of diseases including tumors, infectious diseases, and hereditary diseases. The most critical problem in the clinical applications of an siRNA lies in the fact that an siRNA should be delivered specifically and efficiently to a target cell in vivo. For example, a delivery method is known in which an siRNA is delivered in vivo by means of a high-pressure and high-volume intravenous injection of a synthetic siRNA, utilizing a viral vector. This method, however, uses a viral vector, and a restriction is imposed on the clinical applications of such method from the viewpoint of safety and the like. Consequently, various non-viral systems have been developed that can deliver an siRNA in vivo to the liver, tumor, or other tissues.
Examples of recently developed non-viral delivery systems include the ones using: a cholesterol-siRNA complex (non-patent document 1); stable nucleic acid lipid particles (SNALP) (non-patent document 2); interfering nanoparticles (iNOP) (non-patent document 3) and the like. Among these non-viral delivery systems, SNALP has brought about a great improvement in that the use of SNALP for injecting a clinically appropriate amount of siRNAs has enabled the knockdown of a target mRNA in the liver. However, a therapeutic amount (2.5 mg/kg) of SNALP caused a significant damage to the liver in a crab-eating monkey, whose transaminase level (ALT and AST) exceeded 1000 U/L 48 hours after the administration. Further, a serious disadvantage of SNALP and iNOP is that these delivery systems can only passively transfer an siRNA complex to the liver by utilizing lipophilic nature of the particles that could contribute to the toxicity.
Recently, new types of non-viral delivery systems have been reported, including an siRNA vector (RVG-9R) that can transfer an siRNA via a receptor (non-patent document 4) and “Dynamic PolyConjugate™” (Mirus) (non-patent document 5). The above-mentioned RVG-9R is a short peptide derived from a glycoprotein of rabies virus, added with 9 arginine residues. By utilizing this RVG-9R, an siRNA can be transferred to a nerve cell via an acetylcholine receptor. Meanwhile, the Dynamic PolyConjugate contains a membrane-active form of polymer to which an N-acetylgalactosamine (NAG) is bound, as a ligand targeting a hepatic cell. While the use of these receptor-mediated delivery systems such as RVG-9R and Dynamic PolyConjugate can improve efficiency and specificity of an in vivo siRNA delivery to a target cell, artificially synthesized vector molecules used in these systems still possess hazardous nature that could cause serious side-effects particularly when the dose is increased.
Most recently, a delivery method has been reported, which comprises extracting endogenous lipoprotein, allowing LDL and HDL ex vivo to take in an siRNA to which a cholesterol molecule in the lipoprotein is bound, and introducing the siRNA to the liver via a lipoprotein receptor (non-patent document 6). This complex is taken in by the liver 5 to 8 times more effectively as compared to a free cholesterol-siRNA. However, the complex is effective only to the extent that it can suppress the target gene in the liver by about 55% with a 13 mg/kg intravenous siRNA administration, which is far from sufficient.
Under these circumstances, there have been demands for an siRNA delivery system that can efficiently and specifically deliver an siRNA in vivo and that has a lower risk of causing side-effects.
[Non-Patent Document 1]
    Nature 432:173-178, 2004[Non-Patent Document 2]    Nature 441:111-114, 2006[Non-Patent Document 3]    ACS Chem. Biol. 2:237-241, 2007[Non-Patent Document 4]    Nature 448:39-43, 2007[Non-Patent Document 5]    Proc Natl Acad Sci USA. 104:12982-12987, 2007[Non-Patent Document 6]    Nature Biotechnology 25:1149-1157, 2007